I could have sworn I’d posted about this back when it was in the news a few years ago, but apparently not, and better late than never:

The Berliner Lautarchiv British & Commonwealth Recordings is a subset of an audio archive made between 1915 and 1938 by German sound pioneer, Wilhelm Doegen. Enlisting the support of numerous academics, Doegen sought to capture the voices of famous people, and languages, music and songs from all over the world. The collection acquired by the British Library in 2008 comprises 821 digital copies of shellac discs held at the Berliner Lautarchiv at the Humboldt Universität. It includes recordings of British prisoners of war and colonial troops held in captivity on German soil between 1915 and 1918 and later recordings made by Doegen in Berlin and on field trips to Ireland and elsewhere. The content of the recordings varies and includes reading passages, word lists, speeches and recitals of songs and folk tales in a variety of languages and dialects.

Maev Kennedy, in a 2009 Grauniad story, discusses the British POW angle:

“It’s interesting that there seems to have been no attempt to capture what you might call officer class voices; it was clearly the regional accents that he wanted,” said Jonathan Robinson, curator of social science at the library. “Among the most interesting is the voice from Bletchington – now so close to London it’s barely perceived as having an accent, but I think people would be startled to realise what how West Country the accent of rural Oxfordshire sounded at that time.”
He is particularly fond of the many Yorkshire voices: “That was how my own grandparents would have sounded – but it certainly isn’t how I sound now.”

Thanks, Jeff!


  1. new era 59fifty says

    she has been beat by your ex spouse, in Honest Memo rial Property Cathedral. which in turn afterwards started to be Ginsburg, Mailing Program. He dished up his / her chapel while Fiscal Overseer with the chapel as well as the Child Advancement Heart.
    Jimmy Voss and also the girl
    [spam links deleted, poetic text remains –LH]

  2. > I could have sworn I’d posted about this back when it was in the news a few years ago, […]
    I, too, could have sworn that you had . . .

  3. (1) Your current batch of anti-Bayesian comment spam is poetry.
    (2) The amazing new National Jukebox archive at the Library of Congress ( has a trove of spoken-language recordings from ca. 1900 to 1925. These include dialect comedy routines and political speeches, and for me it was the speeches that came as an enlightening surprise. I expected Woodrow Wilson to speak with a southern accent, for instance, but he didn’t. Score one for the idea of prestige dialects. On the other hand, the Ohioan Warren G. Harding spoke with a regional accent thicker than any I’ve heard in Ohio in my lifetime. Score one more for confirmation of the homogenizing effect of radio.

  4. I can’t help wondering if “Beleuchtung” was somehow triggered by “Bletchington.”

  5. Twenty-five years ago when we lived in Oz I listened to a radio programme of interviews with surviving ANZAC soldiers from the Great War. What struck me was that hardly one had an Aussie accent – they virtually all had regional English accents.

  6. dearie, my great uncle emigrated to Australia after having served at Gallipoli during WW1 next to the ANZACs. I think he had to hold their horses, he was only 16 and keen on riding. Anyway, despite never returning to England he never acquired a proper Australian accent (I only knew him in the 1970s).

  7. I believe the government of New Zealand has an audio archive of speakers born in the 19th century. Interesting because the evolution of the NZ accent can be traced. Unfortunately there is no equivalent for Australia, although there was a project to try to collect data by looking at things like trial transcripts and newspaper stories. I remember a previous discussion but I’m not sure where. It may have been Crooked Timber.

  8. I linked to the Guardian story a couple of years ago at an unrelated thread. AJP commented on it.

  9. I think we commented on the strange pronunciation of ‘father’ (‘fayther’).

  10. Graham Asher says

    I, for one, welcome our new spam overlords – not.

  11. Ah, that will be why I had the déja vu feeling.

  12. John Cowan says

    the strange pronunciation of ‘father’ (‘fayther’)

    What’s strange is that we don’t all say ‘fayther’, or make it rhyme with ‘gather’, or for that matter ‘fadder’. And what’s even stranger is that almost all of us do say ‘fah-ther’.

    In Middle English the word was fader, where the /a/ was a central (“Romance”) variety, the result of merger between Old English front /æ/ and back /ɑ/. However, in this word /a/ could be pronounced either long or short, though the length in OE had been consistently short. In the varieties that underwent this early lengthening, the /a/ went through the Great Vowel Shift in the usual way, leading to the FACE pronunciation.

    But in those varieties (most of them) that kept the ancestral short vowel, we’d expect father to follow gather into TRAP territory, the usual outcome of ME short /a/. But nobody says that: this word unlike gather went to the BATH side of the lexically unpredictable TRAP/BATH split, and what is most remarkable is that it did so in essentially all non-FACE varieties, even those like North American, Irish, and Scottish English where TRAP/BATH never happened at all!

    Apparently this anomaly results from a second lengthening, unique to father, that happened after the Great Vowel Shift and that caused it to fall in with words like palm, calm where the length was due to the loss of ancestral /l/ (with an intermediate stage of /au/). These words made up the entirely new PALM lexical set, pronounced like BATH in varieties where BATH is not TRAP.

    Independent of all this and taking place much earlier was /d/ > /ð/ in this and a few other words in -ther, such as mother, gather, hither, and together. Murther is obsolete, replaced by Latinized Continental Germanic murder, and ladder has reverted, whereas brother was always fricative.

  13. David Marjanović says

    But nobody says that: this word unlike gather went to the BATH side of the lexically unpredictable TRAP/BATH split

    TRAP/BATH is about as predictable as LOT/CLOTH: BATH happens before voiceless fricatives (bath, glass, grass), /ns/ (dance, trans-) and /nt/ (plant). Father is the only irregularity I can think of right now, so I don’t think it should be considered connected to the TRAP/BATH split at all.

  14. Trond Engen says

    I’ll suggest that father is a result of church pronunciation.

  15. John Cowan says

    Father is the only irregularity I can think of right now

    Luckily, Wikipedia is ahead of both of us. The first group is BATH, the second is TRAP.

    /-ðər/: lather, rather vs. blather, gather, slather

    /-f/: calf, chaff, giraffe, graph, half, laugh, photograph, staff, telegraph vs. Aphrodite, chiffchaff, Daphne, gaff(e), graphic, staph

    /-ft/: aft, after, craft, daft, draft/draught, graft, haft, laughter, raft, rafter, shaft vs. kaftan, Taft

    Word-final /-θ/: bath, lath, path vs. hath, math(s), polymath

    Word-final /-s/ brass, class, glass, grass, pass vs. ass (donkey), bass (fish), crass, gas, lass, mass (amount), morass, sass

    /-sk/: ask, bask, basket, cask, casket, flask, mask/masque, rascal, task vs. Asquith, gasket, mascot

    Word-final /-sp/: clasp, gasp, grasp, hasp, rasp vs. asp

    Word-final /-st/: aghast, avast, blast, cast, caste, contrast, fast, last, mast, past, repast, vast vs. bast, iconoclast, paederast, hast

    Miscellaneous /-st/: bastard, caster, castor, disaster, ghastly, master, nasty, pasteurise, pastime, pastor, pastoral, pasture, plaster vs. Aston, astronaut, castigate, chastity, drastic, pasta, pastel, plastic, raster

    /-ʃ/: (always BATH)

    Word-final /-v/: calve, halve, salve vs. chav, have

    /-mpəl/: example, sample vs. ample, trample

    /-nd/: (always BATH)

    /-nt/: (ad)vantage, aunt, can’t, chant, Grant, grant, plant, shan’t, slant vs. ant, banter, cant, fantasy, mantle, phantom, rant, scant

    Word-final /-ntʃ/ (always BATH)

    /-ns/: advance, answer, chance, chancellor, dance, enhance, France, glance, lance, lancet, prance, stance, trance vs. Anson, cancer, fancy, ransom, romance

    The words castle and fasten went into BATH when they had /st/, and stayed there even when they lost the /t/. Raspberry is similar: the central cluster became /zb/. AusE doesn’t have BATH before /m/ or /n/ except in the shibboleths aunt, shan’t, can’t.

  16. @David Marjanović: Rather is also a canonical part of the T/B split, taking /ɑː/ in RP and /æ/ in GA – although I’ve even heard it with /ɑː/ from some regionally unmarked Americans, particularly in news media. Myself, I have /æ/ in rather and all of Wells’s other BATH words, with the sole exception of /ˈɑːnt/ for aunt (a distribution which is pretty common for youngish New Englanders).

    @Trond Engen: That’s an interesting thought. God seems to have undergone something like that: some unmerged Americans anomalously put it in the CLOTH set, with /ɔː/, and it’s also – with gone – one of only two words that take the mysterious /ɒː/ phoneme in Australia.

  17. John Cowan says

    figuratively and literally “high” pronunciation

    For me Gawd (CLOTH) is solely an expletive, but I also use LOT=PALM for the expletive. The ordinary noun is always LOT=PALM.

    I’ve even heard it take /ɑː/ from some regionally unmarked Americans

    (raises hand)

    Not that I am regionally unmarked, whatever that means. However, even though the vowel is back, it is still short, in accordance with the North American Vowel Length Rule. In druthers ‘preference’ < ‘things that I would rather have’), however, I have STRUT like my fellow countrycritters.

  18. Ha, you caught me in an edit. I decided to take that out because for a lot of the remaining unmerged Americans, LOT/PALM and THOUGHT/CLOTH are no longer distinguished by height, only roundedness and/or backness.

  19. Jonathan D says

    I’d say variation in the TRAP/BATH split is the best known difference between different AusE accents. Dance, plant and graph are all BATH in South Australia and TRAP elsewhere. Castle is TRAP in Victoria and BATH elsewhere. I don’t have BATH before /sp/ either, but I’m not sure what the South Australians do.

  20. Bathrobe says

    crass, gas, lass, mass

    For me, ‘gas’ uses /æ/, but unlike the others it is pronounced long (/gæ:s/). Why is this? And where does it fit into the phonology? Is there such a thing as a long /æ/?

    ‘Castle’ is also TRAP in Queensland. Since part of my boyhood was spent in NSW, I adopted the BATH pronunciation.

    I once (many years ago) used ‘castle’ (BATH) and was reprimanded by some Victorians who told me “Hey, we’re Australian, we say ‘castle’ (TRAP)”

  21. John Cowan says

    Is there such a thing as a long /æ/?

    Sure, that’s the BAD/LAD split, which is long/short in England and Australia and diphthong/monophthong in North America. (I don’t have it; I diphthongize before nasals somewhat, but it’s not phonemic.)

    “Hey, we’re Australian, we say ‘castle’ (TRAP)”

    “Pronunciation fills the hearts of all our country-mennnn! Arise! Arise! Arise!”

    (Or as Jack Aubrey’s midshipman had it, the Americans threw tea into Boston Harbor chanting “No reproduction without copulation.”)

  22. David Marjanović says

    Most of the irregularities have (more) recent loans or syllable boundaries on one side, and the other can probably be declared the regular development. Aunt vs. ant, I think, once was deliberate interdialectal borrowing to avoid homophony.

Speak Your Mind